46 



SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 82 



By computing pressure in this way for a network of stations, 

 weather maps bearing unmistakable resemblance to observed weather 

 maps may be computed in advance. In March, 1929, values of pres- 

 sure were computed for one week in advance for 23 selected sta- 

 tions forming a net over the United States and from these computed 

 values lines of equal pressure departures were drawn. The maps 



Fig. 31. — 8 a. m. pressure. 



thus forecasted are compared with the observed pressure distribu- 

 tion in figure 32. The close resemblance of the two sets of maps is 

 apparent. This degree of accuracy can be obtained, however, only 

 when the meteorological cycles are comparatively steady. It is never- 

 theless the goal toward which research is leading and to which it 

 will undoubtedly attain. 



In April, 1929, a diagram was sent to a number of persons, 

 including the Secretary of the Smithsonian Institution, giving a fore- 

 cast of departures from normal temperature by weeks from April 2 

 to September 3 for New York City and for two other stations. Fig- 

 ure 33 gives a copy of this plot for New York City. The broken 

 curve shows the forecast and the continuous curve shows the ob- 

 served departures from normal. The correlation coefficient for the 

 23 weeks is 0.37 ±0.1 2. This correlation taken alone is inconclusive 

 as to the possil)ility of such forecasts, except in the light of other 

 data indicating its possibility. It is believed that forecasts by months 

 and years are feasible on the same basis and by the same methods, 

 but no prolonged test is yet available. 



If the conclusions presented in this paper are verified and accepted 

 by other research workers, as I feel they must be in time, it will 



